We have compared the rates of recovery of pancuronium-induced neuromuscular blockade after administration of neostigmine during anaesthesia maintained with nitrous oxide and intermittent narcotics, halothane or enflurane. Thirty patients were studied in whom anaesthesia was maintained with 70% nitrous oxide in oxygen with: fentanyl or thiopentone, halothane (0.55-0.65% end-tidal), or enflurane (1.3-1.4% end-tidal). Muscle twitch response was measured using train-of-four stimulation. Pancuronium 3 mg/70 kg was antagonized with neostigmine 2.5 mg/70 kg at 10% spontaneous recovery of the first twitch of the train compared with control. There were no significant differences between the times to 10% spontaneous recovery of the first twitch or the rates of train-of-four recovery after neostigmine when the narcotic and halothane groups were compared. However, enflurane anaesthesia, in comparison with fentanyl, was associated with a significant increase in the time to 10% recovery (57.9 +/- 6.0 min v. 35.4 +/- 7.0 min) and with decreases in the train-of-four recoveries at 4, 5, 10, 20 and 30 min after neostigmine. We conclude that, under the conditions of this study, the antagonism of pancuronium-induced neuromuscular blockade with neostigmine was impaired during enflurane but not halothane anaesthesia.